Natural Language Generation from Structured Data

By Daljeet Virdi (, UIUC), Dileep Pasumarthi (Microsoft Bing, UIUC), and Dickey Singh. This is an updated and edited version of the article we published here in January 2020. Natural Language Generation from Structured Data Over the past few years, rapid improvement in state-of-the-art (SOTA) general language models (BERT, GPT-2, GTP-3, XL-Net, etc. and BERT-forks like […]

The Good, Bad, and Ugly of video for sharing insights

Video is engaging, concise, to-the-point, and great for story-telling It has been shown again and again that video is the fastest way to share information when compared to the alternatives, i.e., static reports, slides, and dashboards.  Video displays non-verbal content well, engages audiences, presents content concisely and in a rich manner, incites action, enhances communication, and […]

Importance of Dashboards and why they suck

Business Dashboards There isn’t a single significant business application that does not share insights via dashboards. Dashboards are a window into the operations and performance of a business. They save time by bringing collective business intelligence to one place and are crucial in making informed decisions, aligning strategies, visualizing metrics, achieving goals, and surfacing anomalies […]

Data- vs. intuition-driven Decisions

Business decisions People in businesses make many decisions, ranging from high-profile decisions that define the future, to ones that define the culture, to many ad-hoc decisions that collectively contribute to how the company performs usually guided by the values it stands for. Decision Types McKinsey categorizes decisions based on risk, familiarity, and decision-making frequency. Company-defining […]

Filing a provisional patent to “Patent Pending”

I am filing a new provisional patent after five years. My most recent patent was before the Alice Corp. vs. CLS Bank ruling, which resulted in software patents to be written carefully following strict technical and legal requirements. My approach is to write the provisional patent draft myself along with the drawings, claims, detailed description, abstract, summary, etc. […]

Deep fakes and 2020 elections

I’m here at the TechCrunch Robotics and AI conference (#TCRobotics) in Berkeley and the topic of deep fakes was being discussed at a panel. The conversation of deep fakes in 2020 US elections was interesting in a panel with Alexei Efros (UC BerKeley) and Hany Farid (Dartmouth College.) AI synthesized fake What we have been […]

Organizations accepting AI and ML papers

ML and AI paper submissions need to be made much more accessible.  Everyone is talking about AI.  They are using scandolous stock photos, but that is because no one can understand your research. 😉 Please do not assume math PhDs and rocket scientists as audiences. If including code and data is not possible because of […]